Learning to Associate: HybridBoosted Multi-Target Tracker for Crowded Scene Present by 陳群元.

Post on 30-Dec-2015

224 views 1 download

Transcript of Learning to Associate: HybridBoosted Multi-Target Tracker for Crowded Scene Present by 陳群元.

Learning to Associate: HybridBoosted Multi-Target Tracker

for Crowded Scene

Present by 陳群元

Outline

• introduction• Related work• MAP formulation• Affinity model• Results• Conclusion

overview

STAGE 1STAGE 2STAGE 3STAGE 4

Introduction

• learning-based hierarchical approach of multi-target tracking

• HybridBoost algorithm-hybrid loss function

• association of tracklet is formulated as a joint problem of ranking and classification

ranking

• the ranking part aims to rank correct tracklet associations higher than other alternatives

classification

• the classification part is responsible to reject wrong associations when no further association should be done

HybridBoost

• combines the merits of the RankBoost algorithm and the AdaBoost algorithm .

adaboost

RankBoost

Related work

• the earliest works look at a longer period of time in contrast to frame-by-frame tracking.

• To overcome this, a category of Data Association based Tracking algorithm

• there has been no use of machine learning algorithm in building the affinity model.

MAP formulation

• Robust Object Tracking by Hierarchical Association of Detection Responses

• ours

MAP formulation v1

• R = {ri} the set of all detection responses

j j

j j j

i i

i i i

MAP formulation v1(cont.)

• tracklet association

MAP formulation v1(cont.)

MAP formulation v2

MAP formulation v2(cont.)

• Inner cost

• Transition cost

MAP formulation v2(cont.)

• With these ,we can rewrite it

Affinity model

• Hybridboost algorithm• Feature pool and weak learner• Training process

Hybridboost algorithm

• Ie.

T1T2

T3

Hybridboost algorithm(cont.)

Loss function

• initial

Strong ranking classifier

weak

Update weight

Updatesample weight

Update weight

weak weak weak

Hybridboost algorithm

Weak ranking classifier

Feature & threshold

Feature & threshold

Feature & threshold

Feature pool and weak learner

Training process

• T:tracklet set from the previous stage

• G:groundtruth track set

Training process (cont)

• For each Ti T, if∈• connecting Ti’s tail to the head of

some other tracklet

Training process (cont)

• connecting Ti’s head to the tail of some other tracklet before Ti which is also matched to G

Ranking sample set

Binary sample set

Training process (cont.)

• use the groundtruth G and the tracklet set Tk−1 obtained from stage k − 1 to generate ranking and binary classification samples

• learn a strong ranking classifier Hk by the HybridBoost algorithm

• Using Hk as the affinity model to perform association on Tk−1 and generate Tk

Experimental results

• Implementation details• Evaluation metrics• Analysis of the training process• Tracking performance

Implementation details

• dual-threshold strategy to generate short but reliable tracklets

• four stages of association• maximum allowed frame gap 16,

32, 64 and 128• a strong ranking classifier H with

100 weak ranking classifiers• Β=0.75• ζ = 0

Evaluation metrics

track fragments &ID switches

• Traditional ID switch:“two tracks exchanging their ids”.

• ID switch : a tracked trajectory changing its matched GT ID

• track fragments:more strict

compare

Best features

• Motion smoothness (feature type 13 or 14)

• color histogram similarity (feature 4)

• number of miss detected frames in the gap between the two trackelts (feature 7 or 9).

Strong ranking classifier output

Choice of β

Tracking performance

Conclusion and future work

• Use HybridBoost algorithm to learn the affinity model as a joint problem of ranking and classification

• The affinity model is integrated in a hierarchical data association framework to track multiple targets in very crowded scenes.

• The end– Thank you

System Architecture

完成度 項目100% Ground truth data (CAVIAR、 TRECUID08)50% User Interface for ground truth50% Ground truth Learning phase 1、 2、 3、 430% Feature Extraction

0% Dual threshold method0% Input data training phase 1、 2、 3、 4